A trend analysis methodology for enhanced validation of 3-D LWR core simulations

This paper presents an approach that is being developed and implemented at PSI to enhance the Verification and Validation (V and V) procedure of 3-D static core simulations for the Swiss LWR reactors. The principle is to study in greater details the deviations between calculations and measurements and to assess on that basis if distinct trends of the accuracy can be observed. The presence of such trends could then be a useful indicator of eventual limitations/weaknesses in the applied lattice/core analysis methodology and could thereby serve as guidance for method/model enhancements. Such a trend analysis is illustrated here for a Swiss PWR core model using as basis, the state-of-the-art industrial CASMO/SIMULATE codes. The accuracy of the core-follow models to reproduce the periodic in-core neutron flux measurements is studied for a total of 21 operating cycles. The error is analyzed with respect to different physics parameters with a ranking of the individual assemblies/nodes contribution to the total RMS error and trends are analyzed by performing partial correlation analysis. The highest errors appear at the core axial peripheries (top/bottom nodes) where a mean C/E-1 error of 10% is observed for the top nodes and -5% for the bottom nodes and the maximum C/E-1 error reaches almost 20%. Partial correlation analysis shows significant correlation of error to distance from core mid-plane and only less significant correlations to other variables. Overall, it appears that the primary areas that could benefit from further method/modeling improvements are: axial reflectors, MOX treatment and control rod cusping. (author)